Job Title: MLOps Engineer – Robotics & Perception Systems
Location: San Jose, CA (Onsite)
Employment Type: Full Time
Interview Mode: Virtual
Role Overview
We are seeking an experienced MLOps Engineer who thrives in real-world robotics environments and can own the end-to-end machine learning lifecycle—from raw data ingestion and labeling through training, evaluation, deployment, and performance monitoring.
In this role, you will support a production perception stack spanning 2D and 3D object detection, semantic and instance segmentation, depth estimation, and multi-sensor fusion across camera and lidar systems. Your work will directly impact every vehicle deployed in the field.
This is a deeply cross-functional role, working closely with perception and autonomy engineers, field operations, platform teams, and external labeling vendors. The work is fast-paced, highly tangible, and tightly coupled to real-world performance.
Key Responsibilities
ML Infrastructure & Data Pipelines
Design, build, and maintain scalable ML data pipelines for 2D/3D detection, segmentation, instance segmentation, and depth estimation.
Develop robust workflows for multi-camera and lidar datasets, including data stored in MCAP format.
Own dataset versioning, metadata management, lineage tracking, and reproducibility systems.
Improve training throughput using distributed training frameworks (Ray, PyTorch Lightning, custom launchers).
Optimize data formats, sharding strategies, and loaders for large-scale vision and lidar datasets.
Data Curation & Quality Management
Build automated tooling for dataset selection, active learning, hard-example mining, and outlier detection.
Create and maintain dashboards and automated checks for dataset health, label quality, class balance, and environmental coverage.
Partner with field operations to prioritize data collection runs and close the feedback loop between field issues and dataset updates.
Labeling Operations
Manage internal annotation teams and external labeling vendors.
Define and evolve annotation standards for camera and lidar perception tasks.
Build QA workflows, reviewer tools, and automated label-consistency checks.
Detect systematic labeling issues and drive process improvements and corrective actions.
Deployment & Model Lifecycle Management
Build pipelines for continuous evaluation using telemetry and logs from vehicles in the field.
Monitor model drift, identify edge cases, and manage regression testing across curated “golden” datasets.
Track on-vehicle performance signals to flag data gaps, performance degradation, or unexpected behavior.
Cross-Functional Collaboration
Work closely with perception engineers on calibration, sensor models, data schemas, and on-vehicle inference constraints.
Collaborate with autonomy and navigation teams to align ML outputs with downstream planning and control requirements.
Partner with the platform team to integrate ML pipelines into core infrastructure.
Engage with fleet operations to review real-world performance and prioritize new data collection initiatives.
Required Qualifications
4–7+ years of industry experience in MLOps, ML infrastructure, data engineering, or applied ML engineering.
Strong Python development skills.
Proven experience building and operating large-scale data pipelines for vision and/or lidar datasets.
Hands-on experience managing cloud infrastructure (AWS EC2, S3, IAM, autoscaling, spot instances).
Familiarity with ML lifecycle and orchestration tools such as MLflow, Weights & Biases, Metaflow, Airflow, Ray, or similar.
Experience managing labeling workflows or working directly with annotation vendors.
Strong debugging instincts across the full ML stack—from data quality issues to training failures and evaluation anomalies.
Preferred / Nice-to-Have Qualifications
Experience with PyTorch, CUDA, and common computer vision / 3D perception libraries.
Exposure to multi-sensor fusion, BEV architectures, or 3D perception pipelines.
Familiarity with MCAP, ROS2, Foxglove, and real-time robotics systems.
Prior experience in autonomous vehicles, industrial robotics, or agricultural robotics.
Background in active learning, automated label-quality scoring, or dataset prioritization.
Experience generating synthetic data or simulator-driven dataset expansion.
Experience building auto-labeling pipelines.